Minimum speed reversion - under normal conditions, the A/P and A/T do not control to a speed which is less than minimum speed for the current flap config (about 1.3 Vs). If the actual speed goes below the minimum speed the underspeed limiting symbol appears in the MCP IAS display and the A/P will then command a speed 5 knots greater than minimum speed.
But this minimum speed reversion is not available after G/S capture.
However,
with a radalt inop, the A/P will disconnect 2 seconds after LOC and GS capture and the A/T will NOT retard at flare.
So we have to assume the A/P kicked out after G/S capture, which would have compounded the rush and confusion on the flight deck whilst reengaging it.
But why did the A/T (which thought it was <27' and thus at flare) retard when it should not have?
During a single channel approach (which they sound like they were doing), the A/T will not flare anyway?
If the P1 radalt suddenly dropped to -8' then some more logic should have come into play, namely if Flare Mode is not armed by 350'agl, both A/P's disengage automatically.
Is it possible that just after G/S capture, 2 secs after to be precise, the A/P and A/T disconnected as they should have done because the P1 radalt was u/s, but as the crew struggled to re-engage the A/P and finish configuring, they forgot to re-engage the A/T (TLA already at idle), with the approach turning into a G/F the new F/O was out of the loop - his eyes not registering the decaying speed, the training captain wanted the aircraft configured and the checklists out of the way - his eyes were in the checklist and double-checking the items on it, by reading the checklist he's actively calling the new F/O's attention
away from the ASI, and suddenly - stick shaker..
100 seconds at approach speeds from 180 down to 100 will take you how far? 4.5 to 5nm... how far is the G/S intercept from the threshold?